OAI/NumberTheory/TwoPoint/Circuits/CircuitGateApproximation.lean

GitHub source file

3 declarations, 0 contain sorry, 0 declared as axioms

A single fixed polynomial approximates an arbitrarily wide AND gate under any supplied finite input law. The sampling and its error bound are explicit, and the polynomial degree is bounded by the number of sampled linear factors.

  1. lemma OAI.TwoPointCorrelations.FiniteLaw.probability_mono line 15
  2. theorem OAI.TwoPointCorrelations.gate_pointwise_failure line 26
  3. theorem OAI.TwoPointCorrelations.exists_and_gate_approximation line 57

    Source documentation

    The probabilistic method fixes all samples simultaneously for the input law, without imposing a condition on that law.

    theorem exists_and_gate_approximation {α : Type*} [Fintype α]
        (ν : FiniteLaw α) (k s : ℕ) (b : α → Fin k → Bool) :
        ∃ x : GateSamplingChoices k k s,
          ν.probability (fun y =>
            sampledAndPolynomial x (fun i => if b y i then 1 else 0) ≠ boolAndValue (b y)) ≤
              (7 / 8 : ℝ) ^ s := by
    …

    Mentions: OAI.TwoPointCorrelations.FiniteLaw, OAI.TwoPointCorrelations.GateSamplingChoices, OAI.TwoPointCorrelations.sampledAndPolynomial, OAI.TwoPointCorrelations.boolAndValue